Largescale Instance Retrieval Computer Vision James Hays Many

Large-scale Instance Retrieval Computer Vision James Hays Many slides from Derek Hoiem and Kristen Grauman

Multi-view matching vs … Matching two given views for depth ? Search for a matching view for recognition Kristen Grauman

Inverted file index • New query image is mapped to indices of database images that share a word. Kristen Grauman

Instance recognition: remaining issues • How to summarize the content of an entire image? And gauge overall similarity? • How large should the vocabulary be? How to perform quantization efficiently? • Is having the same set of visual words enough to identify the object/scene? How to verify spatial agreement? • How to score the retrieval results? Kristen Grauman

Can we be more accurate? So far, we treat each image as containing a “bag of words”, with no spatial information Which matches better? e f z h e f a e a h f e e

Can we be more accurate? So far, we treat each image as containing a “bag of words”, with no spatial information Real objects have consistent geometry

Spatial Verification Query DB image with high Bo. W similarity Both image pairs have many visual words in common. Slide credit: Ondrej Chum

Spatial Verification Query DB image with high Bo. W similarity Only some of the matches are mutually consistent Slide credit: Ondrej Chum

Spatial Verification: two basic strategies • RANSAC – Typically sort by Bo. W similarity as initial filter – Verify by checking support (inliers) for possible transformations • e. g. , “success” if find a transformation with > N inlier correspondences • Generalized Hough Transform – Let each matched feature cast a vote on location, scale, orientation of the model object – Verify parameters with enough votes Kristen Grauman

RANSAC verification

Recall: Fitting an affine transformation Approximates viewpoint changes for roughly planar objects and roughly orthographic cameras.

RANSAC verification

Instance recognition: remaining issues • How to summarize the content of an entire image? And gauge overall similarity? • How large should the vocabulary be? How to perform quantization efficiently? • Is having the same set of visual words enough to identify the object/scene? How to verify spatial agreement? • How to score the retrieval results? Kristen Grauman

Scoring retrieval quality Results (ordered): Database size: 10 images Relevant (total): 5 images Query precision = #relevant / #returned recall = #relevant / #total relevant 1 precision 0. 8 0. 6 0. 4 0. 2 0 0 0. 2 0. 4 recall 0. 6 0. 8 1 Slide credit: Ondrej Chum

What else can we borrow from text retrieval? China is forecasting a trade surplus of $90 bn (£ 51 bn) to $100 bn this year, a threefold increase on 2004's $32 bn. The Commerce Ministry said the surplus would be created by a predicted 30% jump in exports to $750 bn, compared with a 18% rise in imports to China, trade, $660 bn. The figures are likely to further annoy the US, which has long argued that surplus, commerce, China's exports are unfairly helped by a exports, imports, US, deliberately undervalued yuan. Beijing agrees the surplus is too high, but says the yuan, bank, domestic, yuan is only one factor. Bank of China foreign, increase, governor Zhou Xiaochuan said the country also needed to do more tovalue boost domestic trade, demand so more goods stayed within the country. China increased the value of the yuan against the dollar by 2. 1% in July and permitted it to trade within a narrow band, but the US wants the yuan to be allowed to trade freely. However, Beijing has made it clear that it will take its time and tread carefully before allowing the yuan to rise further in value.

tf-idf weighting • Term frequency – inverse document frequency • Describe frame by frequency of each word within it, downweight words that appear often in the database • (Standard weighting for text retrieval) Number of occurrences of word i in document d Total number of documents in database Number of words in document d Number of documents word i occurs in, in whole database Kristen Grauman

Query expansion Query: golf green Results: - How can the grass on the greens at a golf course be so perfect? - For example, a skilled golfer expects to reach the green on a par-four hole in. . . - Manufactures and sells synthetic golf putting greens and mats. Irrelevant result can cause a `topic drift’: - Volkswagen Golf, 1999, Green, 2000 cc, petrol, manual, , hatchback, 94000 miles, 2. 0 GTi, 2 Registered Keepers, HPI Checked, Air-Conditioning, Front and Rear Parking Sensors, ABS, Alarm, Alloy Slide credit: Ondrej Chum

Query Expansion Results … Spatial verification Query image New results New query Chum, Philbin, Sivic, Isard, Zisserman: Total Recall…, ICCV 2007 Slide credit: Ondrej Chum

Recognition via alignment Pros: • Effective when we are able to find reliable features within clutter • Great results for matching specific instances Cons: • Scaling with number of models • Spatial verification as post-processing – not seamless, expensive for large-scale problems • Not suited for category recognition. Kristen Grauman

Summary • Matching local invariant features – Useful not only to provide matches for multi-view geometry, but also to find objects and scenes. • Bag of words representation: quantize feature space to make discrete set of visual words – Summarize image by distribution of words – Index individual words • Inverted index: pre-compute index to enable faster search at query time • Recognition of instances via alignment: matching local features followed by spatial verification – Robust fitting : RANSAC, GHT Kristen Grauman

Lessons from a Decade Later • For Category recognition (project 4) – Bag of Feature models remained the state of the art until Deep Learning. – Spatial layout either isn't that important or its too difficult to encode. – Quantization error is, in fact, the bigger problem. Advanced feature encoding methods address this. – Bag of feature models are nearly obsolete. At best they seem to be inspiring tweaks to deep models e. g. Net. VLAD. James Hays

Lessons from a Decade Later • For instance retrieval (this lecture) – deep learning is taking over. – learn better local features (replace SIFT) e. g. Match. Net – or learn better image embeddings (replace the histograms of visual features) e. g. Vo and Hays 2016. – or learn to do spatial verification e. g. De. Tone, Malisiewicz, and Rabinovich 2016. – or learn a monolithic deep network to recognition all locations e. g. Google’s Pla. Net 2016. James Hays